From aa88a3f96469343da9f5bb121b191ac76b0164d5 Mon Sep 17 00:00:00 2001 From: Pieter Vander Vennet Date: Thu, 29 Feb 2024 14:54:14 +0100 Subject: [PATCH] Some more experimentation --- package.json | 4 ++-- scripts/importscripts/compareWebsiteData.ts | 19 +++++++++---------- src/Logic/Web/LinkedDataLoader.ts | 3 ++- 3 files changed, 13 insertions(+), 13 deletions(-) diff --git a/package.json b/package.json index cccf4e0868..606a312315 100644 --- a/package.json +++ b/package.json @@ -1,6 +1,6 @@ { "name": "mapcomplete", - "version": "0.40.0", + "version": "0.40.1", "repository": "https://github.com/pietervdvn/MapComplete", "description": "A small website to edit OSM easily", "bugs": "https://github.com/pietervdvn/MapComplete/issues", @@ -41,7 +41,7 @@ ], "country_coder_host": "https://raw.githubusercontent.com/pietervdvn/MapComplete-data/main/latlon2country", "nominatimEndpoint": "https://geocoding.geofabrik.de/b75350b1cfc34962ac49824fe5b582dc/", - "jsonld-proxy": "http://127.0.0.1:2346/extractgraph?url={url}" + "jsonld-proxy": "http://cache.mapcomplete.org/extractgraph?url={url}" }, "scripts": { "start": "npm run generate:layeroverview && npm run strt", diff --git a/scripts/importscripts/compareWebsiteData.ts b/scripts/importscripts/compareWebsiteData.ts index 87233d34d6..24ff5def51 100644 --- a/scripts/importscripts/compareWebsiteData.ts +++ b/scripts/importscripts/compareWebsiteData.ts @@ -1,10 +1,9 @@ import fs from "fs" -// import readline from "readline" +import readline from "readline" import Script from "../Script" import LinkedDataLoader from "../../src/Logic/Web/LinkedDataLoader" import UrlValidator from "../../src/UI/InputElement/Validators/UrlValidator" // vite-node scripts/importscripts/compareWebsiteData.ts -- ~/Downloads/ShopsWithWebsiteNodes.csv ~/data/scraped_websites/ -/* class CompareWebsiteData extends Script { constructor() { super("Given a csv file with 'id', 'tags' and 'website', attempts to fetch jsonld and compares the attributes. Usage: csv-file datadir") @@ -28,19 +27,20 @@ class CompareWebsiteData extends Script { tags = tags.replace(/""/g, "\"") const data = JSON.parse(tags) - const website = data.website //this.urlFormatter.reformat(data.website) - if(!website.startsWith("https://stores.delhaize.be")){ - return false - } + try{ + + const website = this.urlFormatter.reformat(data.website) console.log(website) const jsonld = await this.getWithCache(cachedir, website) - console.log(jsonld) if(Object.keys(jsonld).length === 0){ return false } const diff = LinkedDataLoader.removeDuplicateData(jsonld, data) - fs.appendFileSync(targetfile, id +", "+ JSON.stringify(diff)+"\n") + fs.appendFileSync(targetfile, id +", "+ JSON.stringify(diff)+"\n\n") return true + }catch (e) { + console.error("Could not download ", data.website) + } } async main(args: string[]): Promise { @@ -66,7 +66,7 @@ class CompareWebsiteData extends Script { handled ++ diffed = diffed + (madeComparison ? 1 : 0) if(handled % 1000 == 0){ - // console.log("Handled ",handled," got ",diffed,"diff results") + console.log("Handled ",handled," got ",diffed,"diff results") } } catch (e) { // console.error(e) @@ -77,4 +77,3 @@ class CompareWebsiteData extends Script { } new CompareWebsiteData().run() -*/ diff --git a/src/Logic/Web/LinkedDataLoader.ts b/src/Logic/Web/LinkedDataLoader.ts index 5c317bccdf..3ff73c83d0 100644 --- a/src/Logic/Web/LinkedDataLoader.ts +++ b/src/Logic/Web/LinkedDataLoader.ts @@ -48,7 +48,8 @@ export default class LinkedDataLoader { ] private static ignoreTypes = [ - "Breadcrumblist" + "Breadcrumblist", + "http://schema.org/SearchAction" ] static async geoToGeometry(geo): Promise {